Advances in Artificial Intelligence for Data Visualization: Developing Computer Vision Models to Automate Reading of Data Plots, with Application to Regression Diagnostics

This work focuses on creating a reliable and consistent way to evaluate regression models using residual plots, which are key tools for checking how well these models work. While statistical tests are often used, they have limits—they only detect specific issues and can sometimes be too sensitive. Residual plots, on the other hand, provide a visual way to spot problems, but they rely on human judgment, which can make them hard to use at scale. This research aims to solve that problem by developing a method to automate the process, making it more efficient and widely usable.